Free 3D Modeling program from Autodesk

[devers6]

Autodesk 123D - Free 3D Modeling Software, 3D Models, DIY Projects, Personal Fabrication Tools

Autodesk has a beta of a new 3D program which is free for now (and supposedly free after beta as well). It looks to be a cross between the free Google SketchUp and parametric apps such as Solidworks and Inventor, and it is designed to create output ready to be made by rapid prototyping via .STL export.

Since it is in beta, it would be great to see if we could convince them to also include simple 2.5D CAM-type features as well, such as outline extraction for 2.5D toolpaths from model features. If they could add some basic intelligence such as tool information and machining ops in the model tree, it could make G-Code conversion much simpler and more conducive to revision. For example, if a hole feature had an attribute that allowed it to be recognized as something appropriate for a drill operation, it would be dead-simple to produce G-Code that would automatically update if the hole was moved.

Or, I could see clicking on a chamfer feature and entering a tool diameter for a bevel cutter; and a toolpath could be inferred at the appropriate Z level and X-Y offset to cut the chamfer efficiently, instead of simply outputting a 3D solid that you would have to "deconstruct" in a separate CAM program to build the appropriate machining ops. Since they seem to be orienting this product towards creating models destined to be easily made, perhaps they might be receptive to building in some extra capability that would make simple CAM operations much easier and more revision-friendly.

A big barrier to entry for hobby (or even commercial) CNC is the requirement to get proficient in a whole bunch of software - CAD to draw your part, exporting in the appropriate format, importing to CAM, producing G-Code, and then bringing it into the CNC control. It sure would be nice if we could condense a lot of those functions into one program that would make it easy to change things without having to start over.

I have downloaded and installed it, but haven't played with it much yet; but it looks promising. And for free, it's definitely worth checking out.

[ger21]

Seems like it's a combination of Solidworks and Sketchup.
I'm going to be playing with it for a bit and see how it does.

[PEU]

very interesting, hoping it remains free after beta, now we need an equally capable and simple CAM application


Pablo

[ger21]

It appears to be designed to export to .stl, so it should work great with Cut3D, MeshCAM, or DeskProto.

[devers6]

This was the point I tried to make in my initial post. If the design program could be extended in just minimal ways, the need for a separate CAM program could be removed. Generating G-Code would be simply a matter of a different kind of Export.

The whole basis of these kinds of programs is to capture "design intent". The designer builds the model using a series of constraints to make sure that the key features of the model are retained, even when modifications are made later on. For example, if a hole needs to be a certain distance from an edge no matter what size the part, the designer puts a constraint on it that keeps the hole at exactly that distance when other dimensions change. The "Feature Tree" of the program helps to manage these constraints by defining the order of operations to perform on the part so that "design intent" is maintained.

The problem is that it simply stops there. "Design Intent" allows the model to be managed so that it absorbs evolution, but it does nothing to assist in preserving "Manufacturing Intent" - the constraints that are needed to be able to transform the virtual model into a physical part.

That is why we have two sets of programs - CAD (computer aided "design") and CAM (computer aided "manufacturing"). Part of it is that the two fields of engineering are different enough that the designer couldn't be expected to have the knowledge to correctly define the manufacturing steps to produce the model as designed. He simply hands it off to someone who does have those skills, who uses a completely different set of tools that are efficient at turning the model into a part.

But 123D seems to purposely try to bridge these concerns by targeting the program towards people who want to actually make something that they are designing - thus the .STL output, the partnerships with MakerBot and TechShop, etc. This focus (one I absolutely support and encourage) could really change the workflow of concept to reality, and I am simply hoping to see the program support 2.5D manufacturing as well as it does 3D.

What I would like would be to get a good thread going with ideas of how 123D could be developed, that would really help us "Zoners" who both design and make our parts. I think there is sufficient community knowledge here to provide some good input to the Autodesk folks who really are asking for help in making the product useful. We could then point them to this thread to get a much higher-bandwidth idea stream into the application, rather than having people make random suggestions as is currently happening in the beta product forums.

I think we have an opportunity here to influence this program which looks to really help move high-power CAD down into a reasonable cost with a wider audience. Autodesk says that's what they want to do - let's provide them with the ideas they need to make it better.

[ger21]

I'm sitting in a Casino hotel right now, and if I could bet that 123D will NOT get CAM functions, I'd make a lot more money that I will at this casino.

If the design program could be extended in just minimal ways, the need for a separate CAM program could be removed.

The size of the install is rather staggering, so I'm guessing it would be a monumental task to add CAM functions to 123D. Any CAM program that extracts 2D toolpaths from 3D geometry is at least several thousand dollars (except one, if your persistent ) The reason is that it's not a trivial process.

Just my opinion of course. By all means try to persuade them, though. I just don't see it happening.

[PEU]

It appears to be designed to export to .stl, so it should work great with Cut3D, MeshCAM, or DeskProto.

Tried cut3d and meshcam not so long ago, I wanted to machine an example file for a potential customer, it was a plastic toy wheel.

None of both programs had a circular/spiral finishing option, both were raster type and the wheel looked awful even with a very fine detail...

In any case, cut3d development seems to be stopped about a year ago, and it seems meshcam took it from there, meshcam looks promising, it just need a couple of more options such as adding round stock and some more strategies, besides these minor details the price is right and the program is very user friendly


Pablo

[devers6]

I'm specifically talking here about 2.5D CAM. 3D CAM is obviously too complex for a free program (although both FreeMill and Vector Art 3D Machinist are out there), and except for the hassle of exporting to a 3D file format and importing it into another program (and potentially with loss of intent that has to be regenerated) this is a workflow that likely won't be improved on in a product at this level.

Here's a scenario I encounter all the time. I draw a simple part in AutoCAD. Maybe it has an inside pocket, a few holes of different sizes, and an outside outline. I have to be take care of things like drawing origin, making sure polylines are closed, etc. Then I export this to a DXF. I close AutoCad and open up CamBam. I import my DXF, then set to work defining a set of machining ops for drilling, pocketing, outline roughing and finishing. Then I generate my G-Code and send it to Mach3 for cutting.

Along the way, I find that my outline is a little too big, or one or more holes are in the wrong place. That means I have to go back into AutoCAD, make my adjustments, and make a new DXF. But because there is no guarantee that the entities in my new DXF have any relationship to the ones in the old DXF, I get to start from scratch in CamBam and build my machine ops all over again. Repeat that 3 or 4 times, and I begin to wonder why I got into CNC in the first place.

Now what if I could right click on the design tree in 123D, pick a feature for say a hole, and then open up an extended tab or context menu to be able to enter a set of data appropriate for this feature. For example, if it is a hole less than 1/2", it would default to a drilling operation and I could enter parameters for spindle RPM, peck distance, etc.

If I chose an outline, perhaps the data would be tool diameter, cut depth, feed rate, cut direction, finish allowance, etc. All of these are just small pieces of data that don't have to mean anything to the CAD program. They can just be an arbitrary list of name-value pairs. They just have to be saved in the project file and always associated with the feature. Then, the smart guys like CamBam and SheetCam can create plugins (not necessarily free ones) that take that data and perform their G-Code conversion magic, and we've saved all the steps and all the aggravation.

This is assuming 123D has a plugin architecture - not sure if it does but most things these days do, and it would be smart of Autodesk to do it. Even if it didn't, if the data were stored in the project file or could be exported in a defined manner, having CAM-oriented data associated with the features in the CAD program means that you wouldn't necessarily have to start your CAM work over every time something changes.

[ger21]

Here's a scenario I encounter all the time. I draw a simple part in AutoCAD. Maybe it has an inside pocket, a few holes of different sizes, and an outside outline. I have to be take care of things like drawing origin, making sure polylines are closed, etc. Then I export this to a DXF. I close AutoCad and open up CamBam. I import my DXF, then set to work defining a set of machining ops for drilling, pocketing, outline roughing and finishing. Then I generate my G-Code and send it to Mach3 for cutting.

Along the way, I find that my outline is a little too big, or one or more holes are in the wrong place. That means I have to go back into AutoCAD, make my adjustments, and make a new DXF. But because there is no guarantee that the entities in my new DXF have any relationship to the ones in the old DXF, I get to start from scratch in CamBam and build my machine ops all over again. Repeat that 3 or 4 times, and I begin to wonder why I got into CNC in the first place.

I wrote a VBA macro for AutoCAD that exports g-code directly from AutoCAD. I can assign machinong operations to polylines and holes (circles), and they are actually saved in the drawing file. If you have a hole with the operations assigned to it, you can move it anywhere you want, and the operations don't change. You can resize it with the properties windows, and the operations still work fine. once you assign operations to a polyline or circle, you can copy and paste them, and the new ones will have the operations assigned to them.
As far as the origin goes, I have a drawing of my machine bed converted to a block in my AutoCAD template. So I just position the part exactly on the machine bed where I want it. The block is ignored by the macro.

As for pocketing, it only does round pockets, but it's not that difficult to draw pocketing toolpaths using the offset command.

To me, making sure polylines are open or closed is trivial, and once you get used to drawing your parts a certain way, shouldn't really be an issue. PEDIT is your friend. As for cut direction, I use a free LISP routine that reverses the direction of a polyline with a single click.

Of course, it's not nearly as full featured as CamBam, but I do all my 2D cutting with it, even though I have pretty good CAM software at my disposal. (Vectric's Aspire).

There's a thread on it here in the AutoCAD section with more info, and here's a link to my webpage for it.
The CNC Woodworker

I have a lot of ideas to improve it, but just don't have the time. Someday.

[mcphill]

And BobCAD...